# Gemini Search Scraper - AI Answers, Citations & Sources (`topapi/gemini-search-scraper`) Actor

Ask Google Gemini your prompts signed out: the full answer, every cited and searched source, cited sentences with offsets, model, location, topics and follow-up suggestions. Brand and domain tracking, follow-ups, repeats, 40+ countries, monitor mode. $0.22 per 1,000 answers.

- **URL**: https://apify.com/topapi/gemini-search-scraper.md
- **Developed by:** [Top Api](https://apify.com/topapi) (community)
- **Categories:** AI, SEO tools, Automation
- **Features:** Limited permissions
- **Stats:** 2 total users, 1 monthly users, 100.0% runs succeeded, 0 bookmarks
- **User rating**: No ratings yet

## Pricing

from $0.22 / 1,000 gemini answers

This Actor is paid per event. You are not charged for the Apify platform usage, but only a fixed price for specific events.

Learn more: https://docs.apify.com/actors/running/actors-in-store.md#pay-per-event

## What's an Apify Actor?

An Actor is a serverless cloud program that runs on the Apify platform. It has two run modes.
In Batch mode, an Actor accepts a well-defined JSON input, performs an action which can take anything from a few seconds to a few hours,
and optionally produces a well-defined JSON output, datasets with results, or files in key-value store.
In Standby mode, an Actor provides a web server which can be used as a website, API, or an MCP server.

Apify vocabulary and the platform model are defined once, in the agent quickstart at https://apify.com/agents.md.

## How to integrate an Actor?

If asked about integration, you help developers integrate Actors into their projects.
You adapt to their stack and deliver integrations that are safe, well-documented, and production-ready.

Do not guess an integration path. Every one of them is in the agent quickstart at https://apify.com/agents.md: the Apify MCP server, Agent Skills with the Apify CLI, the JavaScript and Python clients, the REST API, and the account-free path for an agent with no human to sign in. It also carries the rule on stating cost before the first paid run.

For examples already wired to this Actor's own input schema, see the [API](#api) section below.

Each client library has reference documentation the quickstart does not restate: [JavaScript/TypeScript](https://docs.apify.com/api/client/js/docs.md) (`npm install apify-client`) and [Python](https://docs.apify.com/api/client/python/docs.md) (`pip install apify-client`).

# README

## Gemini Search Scraper - AI answers, citations and sources

Ask **Google Gemini** your prompts the way a new, signed-out visitor asks on gemini.google.com, and get back what Gemini
wrote: the **full answer** (Markdown and plain text), **every page it cited or searched** (position, URL, domain, title,
site name, snippet), **which sentence each page backs** (with its position in the answer), the **model**, the
**location** Gemini placed the visitor in, **topics** and **follow-up suggestions**.

Built for AI-visibility (GEO) work: list your **brands** and **domains** and every answer says whether they were
mentioned, ranked or cited. Ask **follow-ups** in the same conversation, **repeat** a prompt to see how much Gemini varies,
ask from **40+ countries**, and turn on **monitor mode** to see what changed since your last run.

No Google account, no API key, no browser.

### What a run returns

One dataset row per answer (each follow-up turn is its own row). Filled share measured on our runs of 2026-11-15/16
(2,250 answers to shopping, software and services prompts, US):

| Group | Fields | Filled |
|---|---|---|
| Prompt | `query` (your prompt), `prompt` (what this turn asked), `turn`, `repeat`, `country`, `language` | 100% |
| Answer | `answerMarkdown` (headings, lists, tables), `answerText` (plain), `answerWordCount`, `answerLength` | 100% |
| Sources | `sources[]`: `position`, `url`, `domain`, `title`, `siteName`, `snippet`, `favicon`, `section` (`inline` = cited, `searched` = read but not cited), `citedInAnswer` | 99% of answers have sources (creative prompts are answered without a search); url/domain/title/position 100%, favicon 99%, siteName and snippet 40-41% (Gemini gives them for cited pages only) |
| Cited sentences | `citations[]`: `text`, `startIndex`, `endIndex` (position in `answerMarkdown`), `sourceUrls` | 93% of answers; 6,292 of 6,293 offsets slice back to the exact sentence (last 1,000-answer run) |
| Summary | `sourceCount`, `citedSourceCount`, `citationCount`, `citedDomains`, `allDomains`, `webSearchUsed` | 100% |
| Gemini | `model` (e.g. "3.5 Flash-Lite"), `answerLanguage`, `geminiLocation` ("Virginia, USA"), `geminiCountry`, `topics` (top 5 with scores), `followUpSuggestions` | 100%; follow-up suggestions 58% (Gemini offers them on most answers, not all) |
| Tracking | `brandMentions[]`, `brandsMentioned`, `brandsCited`, `domainCitations[]` | when you list brands / domains |
| Monitor | `changeType` (new / changed / unchanged), `newCitedDomains`, `droppedCitedDomains`, `newBrandMentions`, `droppedBrandMentions`, `previousScrapedAt` | when `monitorName` is set |
| Run | `conversationId`, `responseId`, `responseSeconds`, `egress`, `residentialCharged`, `billed`, `error`, `scrapedAt` | 100% |

Three ready-made views: **Answers** (one row per answer), **Sources (one row per page)** and **Brand tracking** (one row
per brand per answer). Pick them in the Output tab or with `?view=sources` on the dataset items API.

### Input

| Field | What it does | Default |
|---|---|---|
| `queries` | Prompts, one per line. Each is one answer. Also accepted: `prompts`, `questions`, `keywords`, `query`, `prompt`. | a 5-prompt sample |
| `country` | Two-letter code. Empty or `us` = a US address at the base price. Any other country asks through a residential address there (see Pricing). | US |
| `language` | Gemini's interface language (`en`, `de`, `pt-BR`...). Gemini usually answers in the prompt's language; `answerLanguage` shows which. | `en` |
| `maxItems` | Stop after this many answers in total, follow-ups and repeats included. You are never charged for more. | every prompt |
| `followUps` | Asked after each prompt, in order, in the same conversation ("which of those is cheapest?"). Each answer is a row. Up to 10. | none |
| `repeats` | Ask each prompt N times, each in a new conversation (1-20). | 1 |
| `trackBrands` | Brand names; add aliases and the brand's domain after commas: `HubSpot, hubspot.com`. | none |
| `trackDomains` | Websites to find among the sources: `example.com`. | none |
| `monitorName` | Compare every answer with the last run of the same name. | off |
| `resetMonitor` | Forget what the named monitor remembered. | false |
| `maxConcurrency` | Prompts asked at once. | 8 |
| `proxyConfiguration` | Not needed. Your own proxy URLs are used as given; Apify RESIDENTIAL counts as an answer from another country. | the actor's own route |

### How to use it

**Is my brand in Gemini's answer?**

```json
{
  "queries": ["best crm for small business", "best crm for real estate agents"],
  "trackBrands": ["HubSpot, hubspot.com", "Salesforce, salesforce.com", "Zoho CRM, Zoho, zoho.com"],
  "trackDomains": ["hubspot.com", "g2.com", "reddit.com"]
}
```

Each answer gets `brandMentions` (mentioned, mention count, first position, rank among your brands, own page cited,
cited sentences), `brandsMentioned`, `brandsCited` and `domainCitations` (cited, only searched, first position).

**A weekly check that says what changed:** schedule the same input with `"monitorName": "crm-weekly"`. Every answer is
delivered (and charged) each run; each row says `new`, `changed` or `unchanged` and lists the cited domains and tracked
brands that appeared or dropped out since the last run.

**A conversation:** `"followUps": ["which of those is cheapest?", "summarise that in one sentence"]` asks both after
every prompt in the same Gemini conversation (rows `turn` 2 and 3).

**How stable is the answer?** `"repeats": 5` asks each prompt five times in fresh conversations.

**What Gemini tells people in Germany:**

```json
{ "queries": ["beste CRM Software für kleine Unternehmen"], "country": "de", "language": "de" }
```

**Every source on its own row:** open the **Sources** view, or call
`https://api.apify.com/v2/datasets/<datasetId>/items?view=sources&format=csv`.

### Pricing

Pay per answer, no start fee:

| Event | Price | When |
|---|---|---|
| `answer` | **$0.00022** ($0.22 per 1,000) | each answer delivered, follow-up turns and repeats included |
| `residential-answer` | **$0.0018** per answer, on top | only when you ask for a country other than the US (or pick Apify RESIDENTIAL) |

Prompts Gemini did not answer are never charged: they come back as a row with `billed: false` and the reason in `error`.
`maxItems` and the run's maximum charge are honoured exactly.

### Good to know

- **Signed-out Gemini.** You see what a new visitor sees: the model Google gives signed-out visitors ("3.5 Flash-Lite"
  on all 2,250 of our test answers) with Google Search grounding. Prompts are text only; an image request came back as a
  text answer.
- **Gemini places you by IP.** US runs go out from US addresses, and an answer Gemini wrote for another country is asked
  again, never delivered as yours. `geminiLocation` shows the city or state Gemini used.
- **Answers vary.** The same prompt can cite different pages from one run to the next; `repeats` and monitor mode measure
  that.
- **Speed.** Gemini takes about 5-20 s to write an answer; 8 prompts are asked at once, about 0.5-0.7 answers a second
  (1,000 answers in 25-30 minutes). A run stops cleanly before its time limit (2 h by default) with what it has.
- **Shopping cards.** Signed-out Gemini returned no product cards on any of our 1,250 test answers (shopping prompts
  included), so the actor has no products field.
- **Restarts are safe.** If Apify moves the run to another server, answers already delivered are neither asked nor charged
  again.

### Output example

A real row from our run of 2026-11-15 with brand and domain tracking on (`answerMarkdown`, `answerText` and the lists shortened):

```json
{
  "type": "answer",
  "query": "best note taking app for photographers",
  "prompt": "best note taking app for photographers",
  "turn": 1,
  "repeat": 1,
  "country": "us",
  "language": "en",
  "answerMarkdown": "For photographers, the \"best\" note-taking app depends heavily on *what* you are trying to capture—whether it's visual mood boards, location scouting data, client shot lists, or camera settings and technical notes. Photographers are typically visual thinkers wh...",
  "answerText": "For photographers, the \"best\" note-taking app depends heavily on what you are trying to capture—whether it's visual mood boards, location scouting data, client shot lists, or camera settings and techn...",
  "answerWordCount": 492,
  "answerLength": 3343,
  "citations": [
    {
      "text": "For photographers, the \"best\" note-taking app depends heavily on *what* you are trying to capture—whether it's visual mood boards, location scouting data, client shot lists, or camera settings and technical notes.",
      "startIndex": 0,
      "endIndex": 213,
      "sourceUrls": [
        "https://evernote.com/learn/best-note-taking-methods-for-photographers"
      ]
    }
  ],
  "citationCount": 8,
  "sources": [
    {
      "url": "https://evernote.com/learn/best-note-taking-methods-for-photographers",
      "domain": "evernote.com",
      "title": "Best Note-Taking Methods for Photographers - Evernote",
      "siteName": "Evernote",
      "snippet": "Each method has its merits, but digital note-taking is often favored for its versatility and efficiency. * Integrating Photography-Specific Notes. Photographers...",
      "favicon": "https://t0.gstatic.com/faviconV2?url=https://evernote.com/&client=BARD&type=FAVICON&size=256&fallback_opts=TYPE,SIZE,URL",
      "section": "inline",
      "citedInAnswer": true,
      "position": 1
    },
    {
      "url": "https://evernote.com/learn/best-note-taking-apps-for-photographers",
      "domain": "evernote.com",
      "title": "Best Note-Taking Apps for Photographers - Evernote",
      "siteName": "Evernote",
      "snippet": "Utilizing tools like Evernote's note-taking capability allows photographers to seamlessly collect and access their thoughts and photographic references all in o...",
      "favicon": "https://t0.gstatic.com/faviconV2?url=https://evernote.com/&client=BARD&type=FAVICON&size=256&fallback_opts=TYPE,SIZE,URL",
      "section": "inline",
      "citedInAnswer": true,
      "position": 2
    }
  ],
  "sourceCount": 6,
  "citedSourceCount": 4,
  "citedDomains": [
    "evernote.com",
    "storyflow.so",
    "zapier.com"
  ],
  "allDomains": [
    "evernote.com",
    "storyflow.so",
    "zapier.com",
    "photographylife.com",
    "medium.com"
  ],
  "webSearchUsed": true,
  "followUpSuggestions": [
    "Want help setting up a photography production template in Notion or Milanote?",
    "How do I set up a photography production template in Notion or Milanote to manage client shoots and shot lists?"
  ],
  "brandMentions": [
    {
      "brand": "Notion",
      "mentioned": true,
      "mentionCount": 4,
      "firstMentionIndex": 1718,
      "cited": false,
      "citedUrls": [],
      "citedSentenceCount": 2,
      "mentionRank": 2
    },
    {
      "brand": "Apple",
      "mentioned": true,
      "mentionCount": 4,
      "firstMentionIndex": 1016,
      "cited": false,
      "citedUrls": [],
      "citedSentenceCount": 1,
      "mentionRank": 1
    },
    {
      "brand": "HubSpot",
      "mentioned": false,
      "mentionCount": 0,
      "firstMentionIndex": null,
      "cited": false,
      "citedUrls": [],
      "citedSentenceCount": 0,
      "mentionRank": null
    }
  ],
  "brandsMentioned": [
    "Notion",
    "Apple"
  ],
  "brandsCited": [],
  "domainCitations": [
    {
      "domain": "reddit.com",
      "cited": false,
      "searched": false,
      "firstPosition": null,
      "urls": [],
      "citedSentenceCount": 0
    },
    {
      "domain": "forbes.com",
      "cited": false,
      "searched": false,
      "firstPosition": null,
      "urls": [],
      "citedSentenceCount": 0
    }
  ],
  "model": "3.5 Flash-Lite",
  "answerLanguage": "en",
  "geminiLocation": "Florida, USA",
  "geminiCountry": "US",
  "topics": [
    {
      "label": "Shopping",
      "score": 0.5873
    },
    {
      "label": "Media / Entertainment",
      "score": 0.2085
    }
  ],
  "conversationId": "c_e51078d2a21d5e27",
  "responseId": "r_950d494bc8a66f8f",
  "responseSeconds": 13.8,
  "egress": "datacenter",
  "residentialCharged": false,
  "changeType": null,
  "newCitedDomains": null,
  "droppedCitedDomains": null,
  "newBrandMentions": null,
  "droppedBrandMentions": null,
  "previousScrapedAt": null,
  "billed": true,
  "error": null,
  "scrapedAt": "2026-11-16T01:45:07.407Z"
}
```

### FAQ

**Do I need a Google account or an API key?** No.

**Which Gemini model answers?** The one Google gives signed-out visitors; `model` shows it on every row.

**Can I ask in another language?** Yes: write the prompt in that language and set `language`; for that country's sources,
set `country` too.

**Why does an answer have sources but no cited sentences?** Gemini searched those pages but did not tie a sentence to
them (`section: "searched"`). Creative prompts (a poem, a joke) are answered without a search: no sources at all.

**How do I get one row per source?** Use the **Sources** view (Output tab) or `?view=sources` on the API.

# Actor input Schema

## `queries` (type: `array`):

What to ask Gemini, one prompt per line, written the way a customer would ask ("best crm for a small agency"). Each prompt is one answer row. Up to 10,000 characters each. Leave empty for a five-prompt sample.

## `country` (type: `string`):

Where Gemini thinks you are (it picks sources and shops by the visitor's location). Empty or us = a US address at the base price. Any other two-letter code asks through a residential address in that country, billed with the residential-answer add-on.

## `language` (type: `string`):

Gemini's interface language (en, de, fr, es, pt-BR, ja...). Gemini usually answers in the language of the prompt; answerLanguage shows what it used.

## `maxItems` (type: `integer`):

Stop after this many answers in total (follow-ups and repeats count). Empty or 0 = every prompt. You are never charged for more.

## `followUps` (type: `array`):

Asked after each prompt in the same Gemini conversation, in order ("which of those is cheapest?"). Each follow-up answer is its own row (turn 2, 3...) and is charged as an answer. Up to 10.

## `repeats` (type: `integer`):

Ask each prompt this many times, each in a new conversation, to see how much Gemini's answer and sources vary. Each answer is charged.

## `trackBrands` (type: `array`):

Brand names to look for in each answer: mentioned or not, how often, first position, rank among your list, and whether a page of the brand is cited. Add aliases and the brand's domain after commas: HubSpot, hubspot.com.

## `trackDomains` (type: `array`):

Websites to look for among Gemini's sources (example.com): cited in the answer, only searched, first position and how many sentences cite it.

## `monitorName` (type: `string`):

Give a name to compare each answer with the last run of the same name: changeType new / changed / unchanged, plus the cited domains and tracked brands that appeared or dropped out. Every answer is still delivered and charged. Empty = off.

## `resetMonitor` (type: `boolean`):

Forget what the named monitor remembered and start again.

## `maxConcurrency` (type: `integer`):

How many prompts are asked at once. Empty = 8 at 128 MB, 12 at 256 MB, 16 above.

## `proxyConfiguration` (type: `object`):

Not needed: the actor picks its own route. Your own proxy URLs are used as given; choosing Apify RESIDENTIAL counts as the residential-answer add-on.

## Actor input object example

```json
{
  "queries": [
    "best crm for small business",
    "best running shoes for flat feet",
    "best project management tool for small teams",
    "best password manager for families",
    "best laptop for college students"
  ],
  "language": "en",
  "repeats": 1,
  "resetMonitor": false
}
```

# Actor output Schema

## `answers` (type: `string`):

No description

## `sources` (type: `string`):

No description

## `brands` (type: `string`):

No description

## `summary` (type: `string`):

No description

# API

You can run this Actor programmatically using our API. Below are code examples in JavaScript, Python, and CLI, as well as the OpenAPI specification and MCP server setup.

## JavaScript example

```javascript
import { ApifyClient } from 'apify-client';

// Initialize the ApifyClient with your Apify API token
// Replace the '<YOUR_API_TOKEN>' with your token
const client = new ApifyClient({
    token: '<YOUR_API_TOKEN>',
});

// Prepare Actor input
const input = {
    "queries": [
        "best crm for small business",
        "best running shoes for flat feet",
        "best project management tool for small teams",
        "best password manager for families",
        "best laptop for college students"
    ]
};

// Run the Actor and wait for it to finish
const run = await client.actor("topapi/gemini-search-scraper").call(input);

// Fetch and print Actor results from the run's dataset (if any)
console.log('Results from dataset');
console.log(`💾 Check your data here: https://console.apify.com/storage/datasets/${run.defaultDatasetId}`);
const { items } = await client.dataset(run.defaultDatasetId).listItems();
items.forEach((item) => {
    console.dir(item);
});

// 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/js/docs

```

## Python example

```python
from apify_client import ApifyClient

# Initialize the ApifyClient with your Apify API token
# Replace '<YOUR_API_TOKEN>' with your token.
client = ApifyClient("<YOUR_API_TOKEN>")

# Prepare the Actor input
run_input = { "queries": [
        "best crm for small business",
        "best running shoes for flat feet",
        "best project management tool for small teams",
        "best password manager for families",
        "best laptop for college students",
    ] }

# Run the Actor and wait for it to finish
run = client.actor("topapi/gemini-search-scraper").call(run_input=run_input)

# Fetch and print Actor results from the run's dataset (if there are any)
print(f"💾 Check your data here: https://console.apify.com/storage/datasets/{run.default_dataset_id}")
for item in client.dataset(run.default_dataset_id).iterate_items():
    print(item)

# 📚 Want to learn more 📖? Go to → https://docs.apify.com/api/client/python/docs/quick-start

```

## CLI example

```bash
echo '{
  "queries": [
    "best crm for small business",
    "best running shoes for flat feet",
    "best project management tool for small teams",
    "best password manager for families",
    "best laptop for college students"
  ]
}' |
apify call topapi/gemini-search-scraper --silent --output-dataset

```

## MCP server setup

```json
{
    "mcpServers": {
        "apify": {
            "type": "http",
            "url": "https://mcp.apify.com/?tools=fetch-actor-details,topapi/gemini-search-scraper"
        }
    }
}
```

The hosted server signs you in with OAuth on first connect, so no API token belongs in this config. Clients without OAuth support can send an `Authorization: Bearer <APIFY_API_TOKEN>` header instead, using a token from API & Integrations in Apify Console (https://console.apify.com/settings/integrations).

## OpenAPI specification

Download the OpenAPI definition: https://api.apify.com/v2/actors/OhAyagDzCsp4VM4TY/builds/NaPPaQDeGNxwC94fm/openapi.json
